Mutational analysis of the bacterial signal-transducing protein kinase/phosphatase nitrogen regulator II (NRII or NtrB)

The signal-transducing kinase/phosphatase nitrogen regulator II (NRII or NtrB) is required for the efficient positive and negative regulation of glnA, encoding glutamine synthetase, and the Ntr regulon in response to the availability of ammonia. Alteration of highly conserved residues within the kinase/phosphatase domain of NRII revealed that the positive and negative regulatory functions of NRII could be genetically separated and that negative regulation by NRII did not require the highly conserved His-139, Glu-140, Asn-248, Asp-287, Gly-289, Gly-291, Gly-313, or Gly-315 residue. These mutations affected the positive regulatory function of NRII to various extents. Certain substitutions at codons 139 and 140 resulted in mutant NRII proteins that were transdominant negative regulators of glnA and the Ntr regulon even in the absence of nitrogen limitation. In addition, we examined three small deletions near the 3' end of the gene encoding NRII; these resulted in altered proteins that retained the negative regulatory function but were defective to various extents in the positive regulatory function. A truncated NRII protein missing the C-terminal 59 codons because of a nonsense mutation at codon 291 lacked entirely the positive regulatory function but was a negative regulator of glnA even in the absence of nitrogen limitation. Thus, we have identified both point and deletion mutations that convert NRII into a negative regulator of glnA and the Ntr regulon irrespective of the nitrogen status of the cell.

PI, acts to control transcription from nitrogen-regulated promoters through the bifunctional kinase/phosphatase nitrogen regulator II (NR,,), the product of glnL (ntrB) (5). Under conditions of nitrogen limitation, NRI1 is autophosphorylated on residue His-139, and these phosphoryl groups are subsequently transferred to the transcription factor NR1, the product of glnG (ntrC) (15,16,18,25,37). Phosphorylation of NR, converts NR, to the form able to activate transcription from nitrogen-regulated promoters (25). Under conditions of nitrogen excess, NR1I and the unmodified form of P,1 act in concert to bring about the dephosphorylation of NR,-P and by so doing prevent the activation of transcription from nitrogenregulated promoters (17,18,25). Nitrogen regulation of transcription is thus due to the control of the availability of P,1 for this interaction by the UT-UR enzyme (5). Several lines of evidence suggest that the NR,-P phosphatase activity resides in NR,, and is elicited by P11. For example, physiology experiments with cells lacking P11 suggest that some of the ability to negatively regulate glnA expression is present, as opposed to Corresponding author. the absence of this capacity in cells lacking NR,, (2,5,8). In this report we present additional genetic evidence suggesting that the phosphatase activity is intrinsic to NR,, and that the role of P,1 is entirely regulatory.
In addition to phosphotransfer from NR,,, there is a second route by which NR1 can become phosphorylated and activate transcription of glnA and the Ntr regulon. Small phosphorylated metabolic intermediates such as acetyl phosphate and carbamyl phosphate can serve as direct substrates for NR, autophosphorylation, and the NR1-P so formed is able to activate transcription of glnA (8). Experiments with intact cells have revealed that the most important phosphorylated metabolic intermediate in this process is acetyl phosphate, since cells lacking the capacity to synthesize acetyl phosphate are unable to activate glnA expression in the absence of NR1, (8).
Alternatively, cells that lack NR11 and retain the capacity to produce acetyl phosphate are unable to negatively regulate glnA expression when the growth medium or genetic background causes an intracellular accumulation of acetyl phosphate. Thus, while NR1, is not essential for the activation of ginA, it is apparently essential for the efficient negative regulation of glnA in response to nitrogen-excess growth conditions (6,8,21). NR, and NR11 are related to many other bacterial proteins that constitute two families of signal transducers known in the aggregate as the two-component systems (reviewed in references 23, 28, and 35). Analysis of the deduced amino acid sequences of the related histidine kinase (kinase/phosphatase) proteins has revealed that these proteins share a conserved domain, usually at the C-terminal end of the protein, and possess in addition unrelated domains. The kinase/phosphatase domain contains three highly conserved regions (23,28,35) (Fig. 1). Region 1, usually contained at the N-terminal end of the kinase/phosphatase domain, contains a conserved histidine residue that is the site of autophosphorylation (14,24). The function of region 2, which contains two conserved asparagine residues, is unknown. Region 3 is a large glycinerich region that can be subdivided into three portions with  (26). The standard single-letter code is used with the following exceptions: X refers to positions at which at least 50% of the kinase family have a nonpolar amino acid (I, L, M, or V), Z refers to positions at which at least 50% of the kinase family have a polar amino acid (A, G, P, S, or T), J refers to positions at which at least 50% of the kinase family have a basic amino acid (H, K, or R), and 0 refers to positions at which at least 50% of the kinase family have an acidic or amidic amino acid (D, E, N, or Q). Wild-card positions with less than 50% conservation among the kinase family are indicated by dashes. The sites at which mutations in NR,, were constructed are indicated by arrowheads. For a more complete depiction of the conserved regions in many kinase/phosphatase proteins, see references 23 and 35. The figure is adapted from Fig. 5 of reference 28. strong homology (regions 3.1, 3.3, and 3.5) separated by two variable spacers (regions 3.2 and 3.4) (Fig. 1). Region 3.5 is apparently involved in nucleotide binding, as shown in the accompanying paper (26).
In the work described in this report, we characterized the phenotypes resulting from alteration of eight of the most highly conserved residues within the kinase/phosphatase domain of NRII. Our results indicate that the negative and positive regulatory functions could be genetically separated and that these highly conserved residues and, indeed, the C-terminal 59 amino acids of NRI, were not required for the negative regulatory function.

MATERIALS AND METHODS
Bacteriological techniques. Media, preparation of plasmid DNA, preparation of competent cells, transformation of cells with DNA, preparation of Plvir phage lysates, P1-mediated transduction, and long-term storage of strains were as described previously (2,22,32). The bacterial strains used in this work are described in Table 1. The plasmids used in physiology experiments (see Table 2) were all similar to the previously described g1nL+ plasmid pgln62 (2,36), which is based on pBR322. The experiment whose results are shown in Table 4 used these and plasmids derived from pACYC184. Both types of plasmids contained a BamHI-HindIII DNA fragment that extends from the middle of the g1nA gene to the beginning of the glnG gene. DNA ligations were performed in low-meltingpoint agarose by using gel-purified bands as described previously (22). Point mutations were constructed by oligonucleotide-driven mutagenesis using the Altered Sites mutagenesis kit (Promega) according to the manufacturer's directions. The DNA sequence of each mutagenized allele was determined completely on one DNA strand, and the sequence containing the mutation and flanking nucleotides was determined on both strands by dideoxy sequencing of double-stranded plasmid DNA, as described previously (13). Sequencing was performed with a Sequenase kit (U.S. Biochemicals) and the set of primers described previously (2). The mutagenic oligonucleotides used in this study are as follows (the site of mutation is   Bat 31 mutagenesis. Plasmid pgln62 was cleaved at the unique Notl site within the glnL gene, and gel purified DNA was treated successively with Bal 31, DNA polymerase I Klenow fragment and deoxynucleoside triphosphates, and T4 DNA ligase as described previously (22), except that all steps wcrc performed within the molten gel slice. The mutagenized DNA was then transformed into strain RB9132R (glnL2001 recAl; Table 1), and transformants were screened for the ability to negatively rcgulate GS on pyruvate-ammonia-glutamine medium by the microassay technique, as described previously (2), and for the loss of the Notl restriction site. Of the 63 transformants analyzed, 3 had lost the Notl restriction site but retained the ability to negatively regulate ginA on pyruvate-ammonia-glutamine medium in the RB9132R background; sequencing indicated that these transformants were a delction of codon 307 (d307), a deletion of codons 307, 308, 309, and 31(0 (d307-3 10), and a large deletion that fortuitously resultcd in a termination codon at codon 291, with all of the preceding sequence being wild type except for a silent mutation in the third position of codon 290 (tcr291). The remaining 60 isolates had lost both the NotI sitc and the ability to negatively regulate ginA and were not cxamined further.
GS assays. In all experiments, we used the transferase assay in the presence of Mn, as described previously (2). Since both adenylylated and nonadenylylated GS are active in this assay, the results reflect the total intracellular concentration of GS. Cells were grown exactly as described previously (2) and were harvested when the culture density reached an optical density at 600 nm of 0.5 (see Table 2). The qualitative GS microassay was performed as described previously (21). Total protein was determined by the method of Lowry et al. (19). All data are the averages of the rcsults from at least two separate cultures grown from separate colonies.
Immunoblotting. An "amplified alkaline phosphatase" kit from BioRad was used for immunoblotting according to the instructions of the manufacturer.

RESULTS
Experimental system for the structure-function analysis of NR,,. We constructed the mutations shown in Fig. 2A and B as described in Materials and Methods. In each case, the mutant allele was inserted into the multicopy cloning vector pBR322 such that the wild-type ginL promoter regulates the various ginL alleles. These constructions created mutant analogs of the previously described glnL+ plasmid pgln62 (36). We then examined the expression of GS (ginA product) in cells of various genetic background containing these plasmids. We also examined the intracellular concentration of NR,, programmed by these plasmids in cells containing a large internal deletion within the chromosomal ginL gene (glnL2001), using the immunoblotting technique.
In wild-type cells, two promoters provide for the expression of glnL, encoding NR,,. When grown on nitrogen-limiting medium, the intracellular concentrations of NR,, NR,,, and GS are each increased as a result of the activation of the chromosomal ginALG operon from the nitrogen-regulated glnAp2 promoter (29). This promoter is activated by NR,-P (25,29).
When grown on nitrogen-excess medium, the glnAp2 promoter is not active (29,30); expression of NR, and NR,, then depends on the ginL promoter (36), which is repressed by NRI-P (5,29 H139N H139V E140Q E140A prestained marked proteins. The polyclonal antibody used cross-reacts with a cellular protein that is not nitrogen regulated; this band serves as a useful internal control and is also shown. served that in 5 Vig of total cell protein, NR,j was barely detectable when the cells had been grown on GNgln medium but was clearly detectable when the cells had been grown on Ggln medium (Fig. 3). As shown, there is at least a 10-fold regulation of the intracellular concentration of NR,, by nitrogen in wild-type cells. Exactly the opposite regulation was seen when NRI, was programmed from only the multicopy pgln62 plasmid, which contains the glnL promoter but lacks glnAp2 (Fig. 3). In this case, cells grown on nitrogen-excess medium had a high intracellular concentration of NR,, (similar to the level seen in nitrogen-starved wild-type cells), but cells grown on nitrogen-limiting medium, which should result in an elevated intracellular concentration of NR,-P, had little NR,, (Fig. 3). As shown, the regulation in this case was also approximately 10-fold (Fig. 3). Thus, cells of the glnL2001 strain containing pgln62 have about a 10-fold-higher NR,, concentration than wild-type cells when grown on nitrogenexcess medium and about a 10-fold-lower NR,, concentration than wild-type cells when grown on nitrogen excess medium. Surprisingly, these variations in the intracellular concentration of NR,, had only a minor effect on the regulation of GS, with the most prominent difference being the more efficient repression of GS in the glnL2001 strain containing pgln62 ( Table 2). We also examined the expression of GS in wild-type cells containing pgln62; as shown, in this background the presence of the plasmid had essentially no effect ( Table 2). Because the pgln62 plasmid restored nitrogen regulation to the glnL2001 strain, we used analogous plasmids containing mutations to discern the effects of the mutations on the positive and negative regulatory functions of NR,,. In order to ascertain whether various mutant alleles resulted in a dominant negative phenotype, we introduced each allele (on a plasmid) into strain YMCIOR, which has an intact Ntr signal transduction system and the capacity to synthesize acetyl phosphate but is recombination deficient because of the recA I allele. To see whether various mutations affected the negative regulatory function of NRII, we introduced the plasmids into strain RB9132R, which is able to synthesize acetyl phosphate, " All plasmids are analogous to pgln62 (glnL+) and contain the indicated mutations, as described in the text. ter indicates a termination codon at the indicated codon, and d indicates deletion of the indicated codon or codons.
GNgln medium contains in addition ammonium sulfate (0.2%, wt/vol) and is nitrogen excess. PNgln mcdium contains pyruvate (01.4%, wt/vol) as the sole carbon source and ammonium sulfate and glutamine as nitrogen sources (each 0.2%, wt/vol) and is nitrogen excess. is recAl, and contains the chromosomal glnL2001 mutation. Previous results had shown that in this strain background, growth on pyruvate, causing high intracellular acetyl phosphate levels, results in elevated GS levels even in the presence of ammonia (8) ( Table 2). The ability of plasmid-encoded altered NR,, proteins to negatively regulate GS on pyruvate-ammonia medium in this strain background was then determined. To see if the negative regulatory activity of altered NRI, proteins depended on the P,, protein, as it does in wild-type cells, we also examined the expression of GS in strain BLR, which is able to synthesize acetyl phosphate and is recAl glnL2001 AglnB (glnB encodes the P,1 protein). To most clearly assess the positive regulatory function of plasmid-borne mutant glnL alleles, we introduced the plasmids into strain WS6005, which is recAl glnL2001 and lacks the capacity to produce acetyl phosphate because of the deletion of pta and ackA (8). This strain is unable to activate glutamine synthetase under any conditions because it lacks the capacity to form NR,-P (8), and we examined whether the introduction of plasmids encoding altered NR,, proteins restored the ability to activate GS on nitrogen-limiting medium. Finally, we further examined the positive regulatory function of NRI, by examining whether plasmids encoding altered NR,, proteins could complement strain RB9132R for growth on the poor nitrogen source arginine. Previous results have indicated that the activation of GS requires only a modest intracellular concentration of NR,-P but that growth on arginine as the sole nitrogen source requires a high intracellular concentration of NR,-P (27).
Thus, this complementation assay is a more stringent assay of the positive regulatory function of NR,, than the measurement of GS expression (27). The H139N, E140Q, and E140A mutations resulted in altered NRI, proteins that are transdominant negative regulators of GS synthesis. When the 13 mutant plasmids were introduced into the YMC1OR background (wild type except recAl), most had essentially no effect on nitrogen regulation ( Table 2). The H-139->N (H139N), E140Q, and E140A constructs were remarkable in that they largely eliminated GS synthesis in nitrogen-limiting medium in the wild-type background; that is, they were transdominant negative regulators. The ter291 construct behaved as if it was a leaky or weak transdominant negative regulator.
Effect of mutations on the PI,-dependent negative regulatory function of NRI,. As noted above, GS is elevated in cells lacking NR,1 (RB9132R) when they are grown on pyruvateammonia-glutamine medium (8). Introduction of the wild-type glnL allele on pgln62 greatly reduced this GS activity in the RB9132R background, which contains P,,, but not in the BLR background, which lacks P1, (Table 2). This reflects the importance of P1, in the negative regulatory function of NR,, ( Table   2). (Note, however, that there was a modest negative regulation of GS synthesis by the wild-type glnL allele even in the BLR strain, which lacks PI,, indicating that negative regulation by NR,, is not entirely dependent on P,,.) Each of the mutant glnL plasmids except the H139V construct was able to bring about the negative regulation of GS in the RB9132R background. (The H139V construct seemed to have a very weak P,,-independent negative regulatory activity that can be discerned only by comparing the expression of GS on Ggln and GNgln media with that of the plasmidless controls.) Since the H139N construct was able to negatively regulate GS, we can state that none of the sites altered in this work are essential for the negative regulation of GS. In the case of the H139N, E140Q, E140A, ter291, and G315A mutations, negative regulation did not require PI, ( Table 2).
Effect of mutations on the positive regulatory activity of NRI,. As shown in Table 2, introduction of wild-type NR,1 (programmed by pgln62) into strain BLR (glnL2001 glnB recAl) resulted in elevated GS levels on the nitrogen-rich GNgln medium. This inappropriate activation of GS expression is apparently due to the positive regulatory activity of NR11 acting in the absence of PI,, as observed previously in similar experiments (5). We examined the expression of GS in the BLR strain background containing the set of mutant plasmids ( Table 2). As shown, the N248D, D287N, G289A, G291A, and dG307 constructs were similar to the wild-type construct in causing the elevated expression of GS on GNgln medium in the strain BLR background. Thus, these mutations do not eliminate the positive regulatory function of NRI. The d307-310 and G313A constructs caused a more modest elevation of GS levels on GNgln medium in this strain background; thus, these constructs are partially deficient in the positive regulatory function of NR11. The H139N, H139V, E140Q, E140A, ter291, and G315A constructs in the strain BLR background resulted in a level of GS lower than that in the plasmidless control in all media tested. These constructs are therefore defective in the positive regulation of GS.
To more clearly assess the positive regulatory function, we examined the effect of introducing the set of mutant plasmids into strain WS6005, which is glnL2001 and cannot produce acetyl phosphate. Cells lacking both NR,, and the capacity to synthesize acetyl phosphate are unable to activate expression of g1nA, and introduction of pgln62 into this background restored the ability to activate GS upon nitrogen starvation (8) ( Table 2). As shown, the N248D, D287N, G289A, G291A, dG307, d307-310, and G313A constructs restored or partially restored the ability to activate GS. These mutations have effects ranging from quite modest (G291A) to clearly observable (d307-310) but do not eliminate the positive regulatory function of NR,, (Table 2). In contrast, the H139N, H139V, E140Q, E140A, ter291, and G315A mutations eliminated the positive regulatory function of NR,1. Either these mutant alleles lack the positive regulatory function entirely or this function is obscured by the negative regulatory function. The results obtained with this strain background mirror the results with the BLR strain background discussed above.
The most rigorous test for the positive regulatory function of NRH is the complementation of the glnL2001 recAl strain (RB9132) for growth on the poor nitrogen source arginine. When this test was performed with pgln62 and the set of mutant plasmids, it was observed that pgln62 and the G289A and G291A constructs clearly complemented the chromosomal glnL2001 mutation and that the dG307 construct gave poor but discernible complementation ( Table 3). None of the other plasmid constructs were able to complement the glnL2001 mutant for growth on arginine as the sole nitrogen source. Thus, the N248D, D287N, G313A, and d307-310 constructs are clearly defective in the positive regulatory function of NR,,.
It should be noted that for the N248D and D287N constructs, this defect is not readily apparent from the GS data (compare Tables 2 and 3). Yet another way to assess the effect of mutations on the positive regulatory function of NR,j is to assess the effect of mutations on the repression of the glnL promoter by NR1. Previous work has indicated that this promoter is repressed by NR,-P (5,36,37); thus, the extent of repression is a measure of the intracellular NR,-P. We examined this repression by directly measuring the relative intracellular concentration of NRI, in RB9132R cells (recAl glnL2001) containing plasmids, using the immunoblot technique (Fig. 3). As shown, those mutations that did not eliminate the positive regulatory function of NRI, (d307, D287N, N248D, G289A, G291A, and G313A) also did not eliminate the repression of NR,, synthesis in nitrogen-limiting Ggln medium, whereas those mutations that eliminated the positive regulatory function (H139N, H139V, E140A, E140Q, and G315A) also eliminated this repression of NRI, synthesis (Fig. 3). The product of the ter291 allele could not be reproducibly observed in immunoblotting experiments; in several experiments it was observed as a very faint band, but in the experiment shown in Fig. 3, it was not observed. Thus, with the exception of those for the ter291 protein, the results of the GS assays and the immunoblotting assays are consistent.
Complementation between mutant glnL alleles. Previous work with the related CheA kinase, part of the two-component regulatory system that controls chemotaxis, has indicated that certain cheA mutations can be complemented by other cheA mutations in recombination-deficient cells (33). Those experiments indicated that the positive regulatory function of CheA is due to separate, compartmentalized functions that can be provided by separate CheA monomers in intact cells. We examined whether the positive regulatory function of NRI, was similarly due to separate compartmentalized functions by examining whether different glnL mutations could complement one another in intact cells. We chose for this analysis the H139V mutation, which causes greatly deficient positive and negative regulatory activities of NR,,, and the d307-310 and G313A mutations, which modestly affect the positive regulatory function but essentially do not affect the negative regulatory function ( Table 2). In order to perform the analysis, these alleles were subcloned into the vector pACYC184, which carries a tetracycline resistance determinant. The abilities of combinations of these alleles to complement were then examined by sequentially introducing pairs of plasmids into strain RB9132R (recAl glnL2001) and examining GS expression. As shown in Table 4, the combination of G313A or d307-310 with H139V resulted in complementation of GS expression. Thus, as with CheA, the positive regulatory function of NR,, seems to be due to separate, compartmentalized functions.

DISCUSSION
We examined the effects of mutations at the highly conserved residues within the kinase/phosphatase domain of NR,,, a member of the histidine kinase family of the two-component regulatory systems. The main conclusion from our work is that, with the exception of the H139V mutation, none of the mutations analyzed had a significant effect on the negative regulatory function of NR,,. Since the alteration of histidine 139 to asparagine resulted in a protein capable of negative regulation (indeed, a transdominant negative regulator), it seems unlikely that histidine 139 is directly involved in the negative regulatory activity. Rather, the introduction of a valine at this position may distort the structure of the protein. Thus, we must conclude from our results that none of the highly conserved sites that we altered has a direct role in the negative regulatory function, that is, the phosphatase activity. Furthermore, since the ter291 allele is a negative regulator of ginA, the C-terminal 59 amino acids of NR,, are apparently not required for the negative regulatory function.
In previous work we isolated and characterized glnL mutations that suppress the Ntr defect resulting from a leaky ginD mutation that decreases the P,,-UT activity and thus results in the inability to convert P,1 to the innocuous P,,-UMP under conditions of nitrogen limitation (2). That selection should have identified mutations affecting the negative regulatory function of NR,,; of 16 mutations that were characterized, 15 mapped either in the nonconserved N-terminal domain of NR,, or in a cluster flanking His-139 (2) (shown for comparison in Fig. 2C). None of the previously isolated mutations that appear to affect the negative regulatory function mapped to the highly conserved residues shared by the histidine kinase proteins of the two-component systems. In the current work we demonstrated that the highly conserved residues shared by the histidine kinases of the two-component systems are not directly involved in the negative regulatory function.
Previous genetic analysis of glnL had indicated that the negative and positive regulatory functions of NR,l can be genetically separated (20,21), and our work extends this conclusion by identifying mutations that result in this separation. For example, His-139 is required for the positive regulatory function, but not for the negative regulatory function, of NR,,. In a number of cases, very highly conserved residues were altered with little effect on glnA regulation. For example, the highly conserved N-248, D-287, G-289, G-291, and G-313 residues were not essential for the positive or negative regu-lation of ginA. These alleles resulted in a subtle defect in positive regulation. Thus, we assume that such alleles of ginL result in altered proteins with slightly diminished kinase activity in vivo.
There is a striking coincidence that the five mutations that appeared to eliminate the positive regulatory function entirely, H139N, G315A, E140Q, E140A, and ter291, also resulted in negative regulation of glnA in the absence of P,,. A current working hypothesis for the role of P,, is that P,, shifts NR,, from a conformation in which it is a positive regulator to a conformation in which it is a negative regulator (5). If this is true, then these mutant NR,, proteins appear to be able to adopt the conformation associated with negative regulation, even in the absence of P,,. One hypothesis for the mechanism of regulation by P,, is that P,, functions to shift the equilibrium between the two conformations of NR,, to that associated with the phosphatase activity.
In the work reported in the accompanying paper, we purified and characterized the H139N, G313A, and ter291 proteins (17,26). It was observed that the H139N and ter291 proteins are unable to be autophosphorylated, while the G313A protein is partially defective in autophosphorylation (26). Furthermore, it was observed that the H139N protein could bind ATP but that the G313A protein was partially defective in the binding of ATP (26). We have also observed that the H139N protein could bring about the dephosphorylation of NR,-P in the absence of P,, (17). The results obtained in those studies thus serve to explain the phenotypes documented in this study.
Finally, in the accompanying report it was demonstrated that autophosphorylation of NRI, occurs by a mechanism in which one subunit in the NR,, dimer binds ATP and phosphorylates the other subunit (26). That result offers an explanation for the intragenic complementation between different ginL alleles documented in this report. Apparently, when H139V and either G313A or d307-310 are coproduced in cells, mixed dimers containing one of each type of subunit can be formed.
The results with purified proteins suggest that it is within such mixed dimers that complementation of the autophosphorylation activity of NR11 occurs (26).